Identification of earthflow-prone slopes (solifluction) in permafrost regions by a combination of satellite-based interferometry and in-situ investigations - a case study in Yushu, Sanjiangyuan, Qinghai-Tibet Plateau

The headwater regions of the Yangtze, Yellow, and Mekong rivers are called Sanjiangyuan in Qinghai-Tibet Plateau. During the last decades, glaciers and permafrost are suffering from rising temperatures and precipitation, thus exacerbating surface instability and fostering landslides consequently. We utilized satellite-based interferometric monitoring to detect instability precursors and reconstruct deformation scenarios with 106 descending Sentinel-1 SAR images acquired from February 2016 to July 2020 in Yushu, Sanjiangyuan region where a typical earthflow occurred. Considering freezing and thawing would induce a large bias from linear deformation, the newly developed model was proposed by integrating in-situ soil temperature and moisture to separate the gravity-driven displacement and seasonal deformation. Four potential landslide prone slopes were identified in a less steep and shady landform with a maximum creep speed up to 45 mm at the regional scale. For the Yushu slope case, slow creep and accelerating creep behaviors were retrieved as precursory with the displacement rate varying from 11 mm/yr to 21 mm/yr before the failure. A seasonal oscillation pattern without gravity displacement was detected at the post-failure stage. In addition, we found that complex piecewise deformation patterns can be characterized by fast uplift (with the maximum deformation up to 20 mm in less than 30 days) in the early winter, and relatively slow subsidence in summer thawing (with the maximum value estimated by 10 mm in more than 37 days). The magnitude and duration of seasonal displacement were highly correlated with the internal hydro-thermal regime, especially soil moisture. Our result highlighted that a deformation separation model is necessary for identifying potential solifluction, evaluating the deformation state, and even forecasting risk in the periglacial regions.